bisaya-kurs korrigiert

backend/scripts/repair-invalid-srs-fragment-sentence-pairs.js

@@ -12,6 +12,7 @@
 import { Op } from 'sequelize';
 import { sequelize } from '../utils/sequelize.js';
 import VocabSrsItem from '../models/community/vocab_srs_item.js';
+import VocabGrammarExercise from '../models/community/vocab_grammar_exercise.js';
 
 function parseArgs(argv) {
   return {
@@ -76,30 +77,84 @@ function isStaleInstructionCard(left, right) {
   return isInstructionLikeText(left) || isInstructionLikeText(right);
 }
 
+function normalizePairSide(value) {
+  return String(value || '')
+    .trim()
+    .toLowerCase()
+    .normalize('NFKC')
+    .replace(/[\p{P}\p{S}]+/gu, ' ')
+    .replace(/\s+/g, ' ')
+    .trim();
+}
+
+function pairSignature(left, right) {
+  return `${normalizePairSide(left)}|${normalizePairSide(right)}`;
+}
+
+function extractLegacyGapFillHints(text) {
+  return Array.from(String(text || '').matchAll(/\(([^)]+)\)/g), (m) => String(m[1] || '').trim()).filter(Boolean);
+}
+
+function extractFixedGapFillHints(text, expectedCount = 0) {
+  const source = String(text || '');
+  const gapRegex = /\{\s*gap\s*\}/gi;
+  const gapMatches = Array.from(source.matchAll(gapRegex));
+  const hints = gapMatches.map((match, index) => {
+    const start = match.index + match[0].length;
+    const nextStart = index + 1 < gapMatches.length ? gapMatches[index + 1].index : source.length;
+    const segment = source.slice(start, nextStart);
+    const hintMatch = segment.match(/\(([^)]+)\)/);
+    return String(hintMatch?.[1] || '').trim();
+  }).filter(Boolean);
+  return expectedCount > 0 ? hints.slice(0, expectedCount) : hints;
+}
+
+async function collectStaleLegacyGapFillPairs() {
+  const exercises = await VocabGrammarExercise.findAll({
+    attributes: ['questionData', 'answerData'],
+  });
+
+  const stalePairs = new Set();
+  exercises.forEach((exercise) => {
+    const qData = exercise.questionData || {};
+    const aData = exercise.answerData || {};
+    const exerciseType = qData.type || '';
+    if (exerciseType !== 'gap_fill') return;
+
+    const answers = Array.isArray(aData.answers)
+      ? aData.answers
+      : (aData.correct ? (Array.isArray(aData.correct) ? aData.correct : [aData.correct]) : []);
+    if (!answers.length) return;
+
+    const text = String(qData.text || '');
+    const legacyHints = extractLegacyGapFillHints(text);
+    const fixedHints = extractFixedGapFillHints(text, answers.length);
+    const fixedSignatures = new Set(
+      fixedHints.map((hint, index) => pairSignature(hint, answers[index]))
+    );
+
+    legacyHints.slice(0, answers.length).forEach((hint, index) => {
+      const signature = pairSignature(hint, answers[index]);
+      if (!fixedSignatures.has(signature)) {
+        stalePairs.add(signature);
+      }
+    });
+  });
+
+  return stalePairs;
+}
+
 async function main() {
   const { apply } = parseArgs(process.argv.slice(2));
+  const staleLegacyGapFillPairs = await collectStaleLegacyGapFillPairs();
   const items = await VocabSrsItem.findAll({
-    where: {
-      [Op.or]: [
-        { learning: { [Op.like]: '%/%' } },
-        { reference: { [Op.like]: '%/%' } },
-        { learning: { [Op.like]: '%?%' } },
-        { reference: { [Op.like]: '%?%' } },
-        { learning: { [Op.iLike]: 'Begrüße %' } },
-        { reference: { [Op.iLike]: 'Begrüße %' } },
-        { learning: { [Op.iLike]: 'Begruesse %' } },
-        { reference: { [Op.iLike]: 'Begruesse %' } },
-        { learning: { [Op.iLike]: 'Drücke %' } },
-        { reference: { [Op.iLike]: 'Drücke %' } },
-        { learning: { [Op.iLike]: 'Druecke %' } },
-        { reference: { [Op.iLike]: 'Druecke %' } },
-      ],
-    },
     order: [['id', 'ASC']],
   });
 
   const matches = items.filter((item) =>
-    isStaleQuestionCueCard(item.learning, item.reference) || isStaleInstructionCard(item.learning, item.reference)
+    isStaleQuestionCueCard(item.learning, item.reference)
+    || isStaleInstructionCard(item.learning, item.reference)
+    || staleLegacyGapFillPairs.has(pairSignature(item.learning, item.reference))
   );
 
   matches.forEach((item) => {

backend/services/vocabService.js

@@ -840,15 +840,14 @@ export default class VocabService {
           const answers = Array.isArray(aData.answers)
             ? aData.answers
             : (aData.correct ? (Array.isArray(aData.correct) ? aData.correct : [aData.correct]) : []);
-          const text = String(qData.text || '');
+          const hintPairs = this._extractGapFillHintPairs(String(qData.text || ''), answers.length);
-          const nativeWords = Array.from(text.matchAll(/\(([^)]+)\)/g), (m) => String(m[1] || '').trim());
 
-          if (!answers.length || !nativeWords.length) {
+          if (!answers.length || !hintPairs.length) {
             return;
           }
 
           answers.forEach((answer, index) => {
-            const nativeWord = nativeWords[index];
+            const nativeWord = hintPairs[index] || '';
             const normalizedAnswer = String(answer || '').trim();
             if (!this._isTrainableSrsPair({ learning: nativeWord, reference: normalizedAnswer })) {
               return;
@@ -867,6 +866,28 @@ export default class VocabService {
     return Array.from(vocabMap.values());
   }
 
+  _extractGapFillHintPairs(text, expectedCount = 0) {
+    const source = String(text || '');
+    if (!source) return [];
+
+    const gapRegex = /\{\s*gap\s*\}/gi;
+    const gapMatches = Array.from(source.matchAll(gapRegex));
+    if (!gapMatches.length) return [];
+
+    const hints = gapMatches.map((match, index) => {
+      const start = match.index + match[0].length;
+      const nextStart = index + 1 < gapMatches.length ? gapMatches[index + 1].index : source.length;
+      const segment = source.slice(start, nextStart);
+      const hintMatch = segment.match(/\(([^)]+)\)/);
+      return String(hintMatch?.[1] || '').trim();
+    }).filter(Boolean);
+
+    if (expectedCount > 0) {
+      return hints.slice(0, expectedCount);
+    }
+    return hints;
+  }
+
   _extractTrainerVocabsFromLessonDidactics(lesson) {
     const vocabMap = new Map();
     const corePatterns = Array.isArray(lesson?.corePatterns) ? lesson.corePatterns : [];
@@ -1913,7 +1934,16 @@ export default class VocabService {
         ['stage', 'ASC']
       ]
     });
-    const validDueRows = dueRows.filter((item) => this._isTrainableSrsPair(item));
+    const validPool = await this.getCompletedLessonVocabPool(hashedUserId, course.id);
+    const validKeys = new Set(
+      (Array.isArray(validPool?.vocabs) ? validPool.vocabs : [])
+        .map((entry) => String(entry?.itemKey || '').trim())
+        .filter(Boolean)
+    );
+    const validDueRows = dueRows.filter((item) =>
+      this._isTrainableSrsPair(item)
+      && (!validKeys.size || validKeys.has(String(item.itemKey || '').trim()))
+    );
     const rows = validDueRows.slice(0, limit);
     const totalDueCount = validDueRows.length;